triumvirate

Orchestrate three adversarial subagents to debate and amend proposed plans.

6|Updated Mar 31, 2026
One-click install
npx skills add https://github.com/Corvalis-LLC/Crow-Stack --skill triumvirate
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: triumvirate
Source: https://github.com/Corvalis-LLC/Crow-Stack/tree/main/skills/triumvirate
Command: npx skills add https://github.com/Corvalis-LLC/Crow-Stack --skill triumvirate

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Solves plan review bottlenecks by orchestrating three adversarial subagents to debate a proposed plan.

Core Features & Use Cases

  • Structured triage and debate by Advocate, Analyst, and Critic to surface strengths, tradeoffs, and risks.
  • Synthesis and amendment mechanism to produce a refined plan and clear next steps.
  • Applicable to evaluating implementation plans, architecture decisions, or significant system changes.

Quick Start

Submit your plan text to be reviewed by the Advocate, Analyst, and Critic for a synthesized amendment and decision.

Frequently Asked Questions about triumvirate

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review an architecture plan to surface hidden risks before implementation?

Architecture plan review can surface hidden risks by orchestrating three adversarial subagents—Advocate, Analyst, and Critic—to debate the proposal, synthesize tradeoffs, and produce an amended plan with clear next steps.

What is the best way to evaluate major system changes and implementation decisions?

Evaluating major system changes is best done through a structured triage and debate process where an Advocate, Analyst, and Critic review the implementation decision to surface strengths, tradeoffs, and risks, followed by a synthesis mechanism that outputs a refined plan.

How do I run an adversarial plan review using multiple subagents?

To run an adversarial plan review, submit your plan text into the workflow. The system applies structured prompts to the Advocate, Analyst, and Critic subagents to debate the proposal, then uses a defined synthesis workflow to generate an amended plan.

Can I use this to review implementation plans for any software engineering project?

Yes, you can use this to review implementation plans across any software engineering project. The adversarial debate applies to evaluating implementation plans, architecture decisions, and significant system changes regardless of the specific project domain.

Does plan review with adversarial subagents require any specific dependencies or frameworks?

Plan review with adversarial subagents requires no external dependencies or specific frameworks. It only requires a clear plan text with structured prompts for the Advocate, Analyst, and Critic roles to execute the defined synthesis workflow.

What format does the output take after a synthesized plan review?

The output of a synthesized plan review is an amended plan accompanied by clear next steps. The synthesis mechanism processes the debate from the Advocate, Analyst, and Critic to produce this refined plan ready for implementation.